AI-powered prediction of HCC recurrence after surgical resection: Personalised intervention opportunities using patient-specific risk factors.

BACKGROUND: Hepatocellular carcinoma (HCC) recurrence following surgical resection remains a significant clinical challenge, necessitating reliable predictive models to guide personalised interventions. In this study, we sought to harness the power of artificial intelligence (AI) to develop a robust predictive model for HCC recurrence using comprehensive clinical datasets. METHODS: Leveraging data from 958 patients across multiple centres in Australia and Hong Kong, we employed a multilayer perceptron (MLP) as the optimal classifier for model generation. RESULTS: Through rigorous internal cross-validation, including a cohort from the Chinese University of Hong Kong (CUHK), our AI model successfully identified specific pre-surgical risk factors associated with HCC recurrence. These factors encompassed hepatic synthetic function, liver disease aetiology, ethnicity and modifiable metabolic risk factors, collectively contributing to the predictive synergy of our model. Notably, our model exhibited high accuracy during cross-validation (.857 ± .023) and testing on the CUHK cohort (.835), with a notable degree of confidence in predicting HCC recurrence within accurately classified patient cohorts. To facilitate clinical application, we developed an online AI digital tool capable of real-time prediction of HCC recurrence risk, demonstrating acceptable accuracy at the individual patient level. CONCLUSION: Our findings underscore the potential of AI-driven predictive models in facilitating personalised risk stratification and targeted interventions to mitigate HCC recurrence by identifying modifiable risk factors unique to each patient. This model aims to aid clinicians in devising strategies to disrupt the underlying carcinogenic network driving recurrence.

Predicting hepatitis B e Antigen seroconversion after pregnancy-The SydPregScore.

BACKGROUND AND AIMS: Achieving Hepatitis B e antigen seroconversion (HBeAg SC) at an earlier age confers a better prognosis. We examined baseline and post-partum factors associated with HBeAg SC after pregnancy. We developed a tool, the SydPregScore, to estimate the likelihood of HBeAg SC in the years after pregnancy. METHODS: A retrospective analysis of an HBeAg-positive pregnant cohort was conducted. Variables including baseline age, parity, alanine aminotransferase level, HBV viral load, quantitative HBsAg, use of antiviral therapy and post-partum flare were collected. Univariate and multivariate Cox regression analyses to determine predictors of HBeAg SC and develop a predictor score were performed. RESULTS: We analysed HBeAg SC rates in 220 pregnancies to 149 HBeAg-positive women from 2006 to 2019. At baseline, their median age was 33 (IQR 29-37), ALT 23 U/L (IQR 17-33) and viral load 8 log10 IU/mL (IQR 6.3-8.2 log10 IU/mL). The majority (133/198, 67.2%) received short-course antiviral therapy to prevent mother-to-child transmission, and 109/192 (56.8%) had a post-partum flare. HBeAg SC occurred in 74/220 (33.6%) after pregnancy (median follow-up 814 days, IQR 405-1531). Multivariate analysis identified baseline viral load <8 log10 IU/mL (HR 2.426 [1.224-4.809], p = .011), baseline ALT ≥2 ULN (HR 2.726 [1.299-5.721], p = .008) and age <35 (HR 2.859 [1.255-6.513], p = .012) to be positive predictors of HBeAg SC. The 'SydPreg Score' estimated the probability of HBeAg SC at 2000 days as 10%, 30%, 70% and 80% for 0, 1, 2, and 3 predictors respectively. CONCLUSION: The SydPreg Score allows the prediction of HBeAg SC in the years after pregnancy. Even in those without elevated ALT, age <35 and viral load <8 log10 IU/mL can identify women with a good chance of subsequent HBeAg SC. Those without a chance may benefit from viral suppression.

Therapeutic drug monitoring in inflammatory bowel disease reduces unnecessary use of infliximab with substantial associated cost-savings.

BACKGROUND: Therapeutic drug monitoring (TDM) of infliximab (IFX) levels in inflammatory bowel disease (IBD) patients can help to guide dose adjustments or changes to therapy for selected patients in remission or with secondary loss of response (LOR). AIMS: To determine how IFX TDM is utilised in a real-life clinical setting and to quantify the potential for TDM to reduce the unnecessary use of IFX. METHODS: Data from all public IBD IFX level testing performed across Australia were prospectively collected from June 2016 to July 2017 to assess physician-reported for testing indications (induction, in remission or LOR) and associated results. The hypothetical influence of IFX TDM was based on an optimal therapeutic range of 6-10 mg/L for mucosal healing. RESULTS: Secondary LOR (reactive TDM) was the most common indication for TDM. These patients have consistently lower median IFX levels: 3.02 mg/L (IQR 1.14-6.67 mg/L) versus 5.22 mg/L (IQR 2.70-8.12 mg/L), P = 0.0001 compared with patients in remission (proactive TDM). TDM helped to identify unnecessary use of IFX in 30.6% of the TDM tests performed in luminal Crohn disease and ulcerative colitis patients, with an associated drug cost saving of $531.38 per IFX TDM test episode. Unnecessary IFX use was identified in 38.9% (96/247) of reactive IFX TDM tests performed and in 19.3% (35/181) of proactive testing. CONCLUSION: Use of both reactive and proactive IFX TDM is cost-effective for IBD management as it informs the clinician where unnecessary use of IFX can be stopped.

Mother-to-child transmission of hepatitis B: Examining viral cut-offs, maternal HBsAg serology and infant testing.

BACKGROUND & AIMS: Antipartum antiviral therapy in the setting of high viral load is recommended to prevent mother-to-child transmission of hepatitis B although recommended viral load cut-offs vary. Quantitative HBsAg has been proposed as an alternative screening strategy to identify high viral load in this setting. Guidelines suggest testing all infants for vaccine response and infection. We set out to re-examine viral load cut-offs; the predictive value of quantitative HBsAg and the need for follow-up infant testing in our cohort. METHODS: A retrospective cohort study of 469 HBsAg positive mother-baby pairs from 2 tertiary hospitals in Sydney was performed. Antiviral therapy (lamivudine or tenofovir disoproxil fumarate) was offered to women with viral load ≥6 log10  IU/mL (high) from 32 weeks gestation. Transmission and vaccine response was analysed according to viral load. The utility of quantitative HBsAg in identifying high viral load was examined. RESULTS: Mother-to-child transmission only occurred in setting of high viral load, in 0.85% (1/117) of those who received antiviral therapy and in 8.66% (2/23) of those who chose not to. Quantitative HBsAg did not accurately identify high-risk mothers HBV DNA ≥6 log10  IU/mL. Successful infant vaccine response was 98.7% overall, and 99.4% when viral load was <6 log10  IU/mL. CONCLUSION: Antiviral therapy initiated at 32 weeks when maternal viral load is ≥6 log10  IU/mL almost completely abrogates transmission. Quantitative HBsAg does not reliably predict high viral load. When maternal viral load is <6 log10  IU/mL, high vaccine efficacy and zero transmission suggests testing infants is of little value.